WordPress Core AI: the MCP adapter goes standalone
Where the WordPress Core AI work stands before 7.0: the MCP adapter moving to a standalone plugin, the connector approval experiment, and a release plan that puts polish ahead of new features.
Tag: AI
Build an AI context layer so your LLM stops forgetting
RAG re-derives everything on every query, so nothing accumulates. This is the Raw and Wiki folder layout, the three control files, and the daily, weekly and monthly jobs I use to keep an AI context layer my agents can read.
WordPress 7.0 AI framework and the connector approval catch
WordPress 7.0 (Armstrong) is past the foundational AI work and into ecosystem plumbing. The connector approval experiment blocks plugins from your API keys until an admin says otherwise, and RC1 lands May 14, so provider plugins need updating now.
Python type annotations that keep data pipelines stable
Dynamic typing is fine in a notebook and expensive in a long-running pipeline. Notes on using type annotations, TypedDict, Literal and union types so a bad value fails at check time instead of four hours into a run.
AI model convergence: why separate models keep agreeing
Vision models and language models look like separate brains, but as they scale their internal structures line up. A look at the Platonic Representation Hypothesis, the pressures behind convergence, and what it means for the data assets you build.
Production-grade AI agents: keep the logic deterministic
Asking an LLM whether a value changed significantly turns a cheap comparison into a slow, untestable guess. Here is the split I use in WordPress: PHP thresholds decide when to act, and the model only writes the message.
Efficient knowledge base design for RAG systems
Dumping PDFs into a vector store is why RAG systems hallucinate and cost too much. This covers curating the source data, chunking by meaning, quantizing vectors, mixing BM25 with vector search through RRF, and testing retrieval with DeepEval.
Long-context foundation models for time-series forecasting
Timer-XL makes the case that decoder-only Transformers beat encoders at forecasting. This looks at how TimeAttention keeps temporal order intact, why patch tokens cut the compute, and what the zero-shot numbers mean for real pipelines.
Multi-Agent Reinforcement Learning for logistics that holds up
MARL for logistics: RL picks the routing strategy, Linear Programming handles packing, and ratio-based observations let one agent run at a rural hub or a metro sorting center. Includes the sequential training loop that keeps agents stable.
Inside WooCommerce for Claude, Automattic’s MCP experiment
WooCommerce for Claude came out of Automattic’s Radical Speed Month. It uses MCP to give the model analytics lookup tables, a store knowledge layer and an AI-readiness score, all on one endpoint. Plus the provider pattern and how to run it locally.